High-resolution view of bacteriophage lambda gene expression by ribosome profiling Xiaoqiu Liua,1, Huifeng Jiangb,1,2, Zhenglong Gub, and Jeffrey W. Robertsa,3 a
Department of Molecular Biology and Genetics and bDivision of Nutritional Sciences, Cornell University, Ithaca, NY 14853
Contributed by Jeffrey W. Roberts, May 23, 2013 (sent for review April 19, 2013)
Bacteriophage lambda is one of the most extensively studied organisms and has been a primary model for understanding basic modes of genetic regulation. Here, we examine the progress of lambda gene expression during phage development by ribosome profiling and, thereby, provide a very-high-resolution view of lambda gene expression. The known genes are expressed in a predictable fashion, authenticating the analysis. However, many previously unappreciated potential open reading frames become apparent in the expression analysis, revealing an unexpected complexity in the pattern of lambda gene function.
strong apparent translation pauses within coding segments (10). There frequently are clusters over translation initiation and termination sites, reflecting slow steps at these stages of translation. As reported by Li et al. (10) and discussed further below, pauses are associated with upstream Shine–Dalgarno sequences. The general expression pattern matches in good detail expectations from decades of detailed study of lambda gene regulation (Fig. 1). All known lambda genes and previously annotated ORFs are expressed during lambda development (except orf206b and probably NP_59778.1), as are the newly identified translated ORFs that we report below.
B
The Uninduced Lysogen. In the uninduced lambda lysogen, cI, rexA, rexB, lom, and bor are the major bacteriophage genes being translated, in agreement with expectation (Dataset S1). Others appear significantly over background, including the early genes ea8.5 and ea59 (both of unknown function), and the immediate early genes N and cro; the latter presumably become transcribed when repression occasionally fails, although there clearly is not enough expression of gene N to allow expression of most of the delayed early set of genes that depend on the gene N transcription antiterminator. A few other genes appear detectably over background (Dataset S1). Two array experiments (3, 4) measuring RNA agreed in identifying in uninduced lysogens most of the set of five that we find, but each report found other genes significantly expressed that we do not find and, in fact, there was little agreement between the two array studies about these others. Differences with our measurements presumably reflect the fact that RNAs are not uniformly translated, in addition to uncertainties of array measurements.
acteriophage lambda is an original and exemplar organism that has guided the discovery of basic genetic regulatory processes, including transcription repression, activation, and antitermination (1, 2). Lambda has provided an important model to understand the interaction of a virus with its host. Programs of lambda gene expression that establish and maintain the prophage state, and that mediate the lytic pathway of growth upon infection or induction, are well understood in terms of the basic biochemical mechanisms and the timing of regulatory protein function. Both the classic, focused analysis of gene expression and the modern global approaches such as array hybridization (3, 4) have been applied to elucidate the complex pattern of gene function during lambda growth. A recent method, ribosome profiling, provides a further detailed and precise view of gene expression by capturing the instantaneous translation sites of all of the ribosomes in a cell (5–8). We have applied ribosome profiling to the process of lytic growth of bacteriophage lambda to map in detail the expression of lambda proteins and to infer unique loci of translation. We, of course, confirm the known patterns of gene expression, but we also expand understanding in several ways, including a complete catalog of gene expression, the discovery of unique functional open reading frames (ORFs), and the discovery of bacterial genes expressed during phage development. Results and Discussion
Overview of Method and Approach. We chose temperature in-
duction of the classic cI857 repressor mutant of lambda in a lysogen of Escherichia coli MG1655 to synchronize the lytic process, sampling the lysogen and control nonlysogen both before and 2, 5, 10, and 20 min after shifting the temperature from 32 °C to 42 °C. The last sample time was chosen to be before any significant cell lysis, but during the later stages of lytic gene expression. Sequencing produced ∼106 ribosome prints per sample, which were mapped onto both phage and bacterial genomes and visualized in the Gbrowse genome browser at EcoWiki (9). Total protected nucleotides within ORFs were summed to determine the density of translation of each reading frame (5). We take this number to indicate the overall rate of translation, although obviously we are assuming that pauses in translation do not excessively affect the overall rate. Because the expression level is not normalized for the copy number of the replicating phage DNA, it thus encompasses both the effect of DNA template availability on mRNA synthesis and the efficiency of utilization of messengers by ribosomes.
Gene Expression During Lytic Growth. After repression is relieved, lambda gene expression occurs in two waves (11, 12). Derepression enables promoters pL and pR to function, providing expression of genes N and cro at the earliest time; these genes are the only two lytic genes highly expressed at 2 min after derepression (Fig. 2 A and B). N is a transcription antiterminator that potentiates transcription of the early genes to the right of N and cro. Early genes are expressed increasingly from 5 to 10 min, and then less at 20 min (Fig. 2 A and B). The decrease in early gene expression in the last interval is attributed to the activity of the lytic repressor Cro, which represses both pL and pR as its concentration builds up in the cell (13). The last of the early genes on the right is Q, which encodes an antiterminator that provides expression of all of the late genes (14). Late gene expression only appears significantly at 10 min, and then increases greatly by 20 min (Fig. 2C), reflecting its dominance in the last period of
Author contributions: X.L. and J.W.R. designed research; X.L. performed research; X.L., H.J., and Z.G. analyzed data; and X.L. and J.W.R. wrote the paper. The authors declare no conflict of interest. Data deposition: The data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE47509). 1
X.L. and H.J. contributed equally to this work.
2
General Features of the Translation Pattern. As shown in more
Present address: Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, 32 XiQiDao,Tianjin Airport Economic Park, Tianjin 300308, China.
detail below, the pattern of ribosome occupancy over individual protein coding sequences follows that previously reported, displaying highly asymmetric distributions of protected sites due to
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1309739110/-/DCSupplemental.
11928–11933 | PNAS | July 16, 2013 | vol. 110 | no. 29
3
To whom correspondence should be addressed. E-mail:
[email protected].
www.pnas.org/cgi/doi/10.1073/pnas.1309739110
lambda gene expression as phage structural proteins accumulate. Fig. S1 and Dataset S1 document in more detail the patterns of early and late gene expression, including translation of newly recognized ORFs in both periods. Previous work has shown that although transcription across the lambda late gene region is approximately uniform, reflecting the single mRNA synthesized for the late genes under influence of the gene Q antiterminator, the yields of various proteins from the late gene transcript are very different (15, 16). This differential translation of late genes is apparent in the ribosome profile (Fig. 2 C and D). It is of interest to compare the instantaneous rate of expression of each transcript to a measure of the ultimate requirement for each structural protein in the phage particle (2, 17). Fig. 2E shows some correlation, in particular for the two most abundant proteins D and E, and for some but not others of the least abundant proteins. It is noteworthy that rexB, a gene expressed in the prophage, shows increased expression at late times, consistent with previous genetic evidence (18). rexB is transcribed from the independent promoter pLIT, and it is likely that the enhanced expression of rexB at late times is due to increased gene dosage as phage DNA is replicated. The Pattern of Ribosome Progression. As reported (10), the distribution of ribosomes across reading frames is highly heterogeneous, displaying strong pause sites that are correlated both with sites of initiation and termination, and with internal Shine– Dalgarno sequences that presumably stall ribosomes by annealing with the end of 16S RNA, as in initiation. To illustrate the general pattern of translation, Fig. 2D shows a display of the distribution of ribosomes over a segment of approximately 11 kB of the late genes. Note that the scales are drastically different for displays of the different time samples in Fig. 2D; thus, there is a difference of ∼100 in scale between 5 and 20 min, but the overall patterns are quite similar. In addition to internal pause sites, there frequently are collections of ribosomes over initiation and termination codons of the reading frame, as noted (10). Fig. S2 uses the analysis of Li et al. (10) to confirm that ribosomes in the coding sequence tend to pause where Shine–Dalgarno sequences are positioned upstream to interact with the ribosome. Liu et al.
Frameshifting. A frameshift occurs between lambda genes G and
T, resulting in a fusion protein between these reading frames when approximately 3.5% of the ribosomes slip back a nucleotide at the “slippery” sequence 5′-GGGAAAG-3′ near the end of the G reading frame (19). There is a distinct concentration of ribosomes over this sequence, and at the normal termination codon of the G reading frame, which is evident upon examining the profiles (Fig. 2F). A typical protected RNA segment covering the slippery sequence (in bold) is TCTGCGGGAAAGTGTTCGACGGT, and a typical segment covering the termination codon of G is TCGAGGGTGAGCTGAGTTTTGCCCT. Translation of a Regulatory RNA. Regulation of late gene expression in lambda and related phages occurs through transcription antitermination, by the product of gene Q, of a constitutive transcript from the late gene promoter (14); in lambda, this RNA (lambda 6S RNA, not to be confused with the cellular 6S RNA) is 200 nt long. An ORF, orf-64, begins within this RNA and extends through the transcription terminator. orf-64 shows ribosome-protected fragments at a low but significant level, ∼5% the level of the adjacent Q gene and ∼10% the level of gene S, the first late gene regulated by Q (Fig. 3A). Although it is clear from biochemical analysis with purified proteins that the Q protein and transcription elongation factor NusA suffice to cause antitermination of transcription at the terminator of 6S RNA (14), the occurrence of concurrent translation in the cell could modulate or enhance the antitermination process. It may be significant that a cluster of ribosomes appears over the upstream half of the intrinsic transcription terminator stem, a configuration that would inhibit termination. Such structure is reminiscent of attenuation control in bacterial operons (20) and could suggest a distant evolutionary relationship between these regulatory mechanisms. Unique Translated ORFs. A notable feature of the ribosome profiles is the abundance of translation in regions of the genome without characterized genes but in potential ORFs of unknown function, many of which have not been annotated (Dataset S2 and Fig. S3). We discuss below selected instances of such translation that seem to be noteworthy. However, we also attempted to catalog all such PNAS | July 16, 2013 | vol. 110 | no. 29 | 11929
GENETICS
Fig. 1. An overview of gene expression measured by ribosome profiling across the lambda phage genome during lambda prophage induction. Bar plots centered on dashed line circles from inside to outside show expression levels by ribosome profiling at different times after shifting the culture to 42 °C: blue, 0 min; cyan, 2 min; green, 5 min; orange, 10 min; and red, 20 min. Annotated genes of lambda are shown on the outside. At each time point, the reads per kilobase per million (RPKM) for lambda genes was normalized by scaling between 0 and 1, with 531 being the maximum at 0 min and 9,091 at 2 min. Because the RPKM values for some genes at 5, 10, and 20 min are too high, the maximum values were set at 10,000, where RPKM is reads per kilobase of coding sequence per million mapped reads, as originally denoted (31).
Fig. 2. Expression patterns of the lambda genome from ribosome profiling. (A–C) Gene expression pattern in the early left operon (A), early right operon (B), and late gene operon (C) at different times after induction of a lambda prophage. Reads were summed over each reading frame. (D) Higher resolution view of the ribosome occupancy profile for some late genes during lambda prophage induction. (E) Correlation of gene expression levels for phage structural genes with the protein copy numbers in the purified lambda particle. (F) Ribosome occupancy profile at the translational frameshift region of lambda genes G and T. The red dashed rectangle indicates the frameshift site. Ribosomes are stalled at the slippery sequence GGGAAAG in the G reading frame; a few of these signals (approximately 3.5%) shift back one base on the mRNA into the T reading frame and continue to the end of T (19). A cluster of ribosomes over the G termination codon also is apparent.
potential ORFs in a systematic way: Dataset S2 lists 55 potential ORFs distinct from known genes and previously annotated ORFs, all of which display significant translation over background. This list was obtained by (i) computing all potential ORFs greater than 5 aa in length with either ATG or GTG initiation and a termination codon (∼3,000); (ii) removing all ORFs with apparent translation levels below a chosen background; and (iii) removing from this list previously recognized genes or ORFs, and completely overlapping ORFs, leaving 55. Some of these potential ORFs display significant upstream matches to the Shine–Dalgarno initiation sequence, but many do not (Dataset S2); however, this feature is not a requirement for significant translation. There is no evidence that any of these ORFs represent functional genes. Furthermore, we cannot rule out the possibility that at least some of the signal, particularly the weakest, reflects an entirely irrelevant background association of RNA with ribosomes. Many potential ORFs are from regions of the genome designated “inessential,” such as the “b2” region and the segment between bor and the right end (2) (Fig. S3). It is clear, however, that inessential in the laboratory context does not mean useless in nature. Furthermore, the level of expression of many of the potential ORFs in Dataset S2 is comparable to that of known genes (as we describe below), consistent with the notion that this list 11930 | www.pnas.org/cgi/doi/10.1073/pnas.1309739110
could include candidates for regions of important translation activity, including potential active polypeptide products. We first consider potential translated ORFs that are coded in the predominant direction of known transcription. In several cases, these ORFs occur between genes of the late region when there is space—a rare occurrence, because throughout the late genes, a termination codon generally meshes closely with or even overlaps the next initiation codon (2). In one case, 150 bp separates the termination of lambda tail gene L from the initiation of tail gene K; an ORF of 76 amino acids (ORF 322) starts 6 bp after the termination of L and proceeds through the first 11 amino acids of K (Fig. 3B). Juhala et al. (21) noted that this ORF is homologous to the beginning of the counterpart of gene K in the lambda-related phage HK022, and they suggested that a frameshift mutation might have severed a gene ancestral to lambda K. Our data shows that ribosome occupancy of ORF 322, largely clustered just after the initiation codon, is comparable to that of adjacent genes. Because it is expected that transcription of the late gene region is uniform, this result means that the translation activity of ORF 322 is significant. Possibly ORF 322 and lambda K together constitute the active protein that corresponds to the related protein of HK022. In a second case, found at the beginning of the late gene transcript, there is an interval of 375 bp between the end of the Liu et al.
previously recognized orf-64 and the first nucleotide of gene S, the first recognized late gene (2). Two additional ORFs occur in this interval: ORF 915 (of 15 aa) and ORF 916 (of 12 aa). The density of translation over ORF 916 is approximately one-half that of gene S (Fig. 3A and Dataset S2). The inessential lambda late gene stf is interrupted by a mutation in laboratory strains, resulting in an N-terminal fragment encoded by orf-401, and a potential C-terminal fragment encoded by orf-314 (22). orf-401 is translated at a level comparable to some late genes, and orf-314 is translated much more weakly. One of two potential ORFs between the fragments (ORF 438) appears to be significantly translated. Thus, there is the potential
for expression of the distal parts of the interrupted stf gene, but no evidence that it has any significance. Between the leftward genes exo and Ea22, there is an interval of 953 bp with several previously recognized leftward-directed ORFs (orf60a, orf63, orf61), as well as orf73 (also ORF 2310 in Dataset S2) identified with a function named bin (23). We identify another, ORF 2313 (31 aa), which is shown in Fig. 3C, which is expressed comparably to the known genes like exo and Ea22. In several cases, there are translated ORFs in regions previously known to be transcribed from both strands. Thus, in the region of predominant leftward transcription to the left of gene N, the gene sieB is transcribed rightward (Fig. 4 A). Mostly overlapping sieB is
Fig. 4. Unique translated ORFs in regions of overlapping or predominant opposite strand transcription. (A) Ribosome occupancy profiles downstream of gene N, with 20 min of induction time. Unique ORFs in this region are shown in green; most are antisense to gene sieB, but in the direction of prevailing transcription from pL. (B) Ribosome occupancy profiles of eight unique ORFs at the end of lambda genome, with 20 min of induction time; these ORFs are antisense to bor and lambda p78 (NP_597781.1) but in the direction of the Q-dependent late gene transcription. (C) Ribosome occupancy profiles of plus strand ORFs in the b2 region, with 20 min of induction time. The 14 ORFs shown are antisense to ea47, ea31, and ea59, but in the same direction as Q-dependent late gene transcription, which may thus converge with the N-dependent transcription from pL.
Liu et al.
PNAS | July 16, 2013 | vol. 110 | no. 29 | 11931
GENETICS
Fig. 3. Examples of expressed unique ORFs in the prevailing direction of transcription. (A) Ribosome occupancy profiles downstream of gene Q, with 20 min of induction time. Some ribosome occupancy was found on orf-64 (red), which extends beyond the 194-nt pR’ transcript that is terminated (in the absence of the gene Q antiterminator) at nucleotide 44780. The sequence of the terminator is shown, including the hairpin (red) and poly U (underlined) sequence at the end. Two other unique ORFs (ORF 915 and ORF 916) in this region also showed significant ribosome occupancy. (B) Ribosome occupancy profiles between genes L and K, including ORF 322, with 20 min of induction time. (C) Ribosome occupancy profiles between ea22 and orf61, with 20 min of induction time.
Fig. 5. Effect of phage induction on E. coli protein synthesis. (A) Increase in E. coli gene translation around the attB site due to escape synthesis. Time at 42 °C: blue, 0 min; cyan, 2 min; green, 5 min; orange, 10 min; and red, 20 min. (B) E. coli genes with functions likely related to lambda phage growth and significantly up-regulated during prophage induction. (C) E. coli genes mostly with known functions not obviously related to lambda phage growth but significantly up-regulated during prophage induction. (D) Phage induction leads to a decrease in ribosome occupancy on E. coli genes, and approximately 30% of total reads map to the phage genome at 20 min. (E) Decrease of E. coli gene translation during lambda induction.
a set of substantially translated ORFs, all oriented leftward, and presumably derived from the N-dependent leftward transcription: ORF2423 (24 aa), ORF 3070 (21 aa), ORF 2426 (20 aa), ORF2429 (37 aa), ORF2432 (30 aa), ORF2433 (26 aa), and ORF2434 (15 aa). Between genes RZ and Nu1 of the rightward-directed late transcript, there is an interval of 2,278 bp that contains the leftward-directed gene bor and extends to the cos DNA packaging site (Fig. 4B). This interval contains the rightward-directed ORF 948 (45 aa), ORF 950 (35 aa), ORF 958 (31 aa), ORF 966 (27 aa), ORF 1830 (24 aa), ORF 975 (9 aa), ORF 976 (28 aa), and ORF 1832 (33 aa). Most of these display substantial translation. There is a concentration of translated ORFs directed counter to the prevailing direction of transcription in the “b2” region. This substantial region of approximately 5 kb, which can be deleted without impairing growth (2) (although prophage integration is compromised), contains three well-recognized “early” genes, ea59, ea31, and ea47; all are directed leftward and presumably expressed primarily via gene N antiterminator-influenced leftward transcription from the early promoter pL. (ea59 is also significantly expressed in the lysogen, so that another promoter also may be present.) Several translated ORFs, ORF 467 (53 aa), ORF 472 (14 aa), ORF 486 (14 aa), ORF 488 (17 aa), ORF 494 (28 aa), and ORF496 (36 aa) are all within ea47 and are directed rightward (Fig. 4C). Several rightward-directed ORFs, ORF 499 (of 21 aa), ORF 1509 (of 5 aa), ORF 503 (of 17 aa), ORF 1510 (of 39 aa), and ORF 507 (of 7 aa), are in the 400-bp intergenic region between genes ea47 and ea31 and are expressed comparably to the three leftward genes (Fig. 4C). Within ea31 are two rightwarddirected ORFs (Fig. 4C). Remarkably, ORF 511 is 73 aa long, and ORF 1522 is 30 aa long. Both are translated comparably to the adjacent leftward genes ea31 and ea59. There also are rightward-directed ORFs within the early region to the right of the attachment site (not illustrated). Thus, within ea8.5, the overlapping ORFs 585 (of 130 aa) and 591 (of 41 aa) show substantial translation. The source of the transcription for these newly recognized rightward ORFs is unknown. It may of course be the gene Q antiterminator-influenced late gene transcription, which is expected to proceed efficiently at least through the tail fiber genes stf and tfa, and could easily extend well into the early region, although its efficiency likely would be reduced through collision with the predominant leftward-directed N-protein-influenced transcription from promoter pL. A more detailed display of expression data of some potential ORFs in shown in Fig. S4. Changes in Host Gene Expression After Lambda Induction. Expression in the host was measured before and after lambda induction 11932 | www.pnas.org/cgi/doi/10.1073/pnas.1309739110
through ribosome profiling of annotated E. coli genes, in the same experiments described above. We find that although approximately 1,000 genes are down-regulated during the 20 min of induction, a substantial number, 120, are up-regulated (Dataset S3). Most of these changes occur primarily late in induction. Approximately one-half of the up-regulated gene expression has a relatively trivial origin: Genes that surround the attachment site attB are replicated along with the prophage (“escape replication”) and are overexpressed primarily because of the increased gene copy number (Fig. 5A). This enhanced expression occurs in a region of 300–400 kB surrounding the prophage integration site (4). No down-regulated genes were found in this region. One well-characterized set of up-regulated genes is the nearby gal operon, overexpression of which depends on both overreplication and gene N-dependent antiterminated transcription from the phage pL promoter (4). Transcription of gal operon genes is enhanced approximately 13-fold in induced lysogens relative to nonlysogens (4); we found that ribosome occupancy of the gal operon genes galE, galT, and galK increased approximately 13–16-fold, in good agreement (Dataset S3). A prominent group of up-regulated genes located far from attB and, thus, stimulated by some metabolic change in the cells, is the set of SOS DNA damage genes (examples in Fig. 5B) (24). Because our induction protocol used temperature inactivation of phage repressor and not explicit DNA damage (e.g., UV irradiation), the induction of SOS genes must be due to some aberration in DNA metabolism that arises during phage growth, likely resulting from the replication of phage DNA; this anomaly might be the accumulation of single-stranded regions and incompletely replicated phage chromosomes, for example. It is noteworthy that increased translation of the SOS repressor LexA occurs along with that of the other SOS genes, although the concentration of LexA must be diminished. This result is in fact as expected, because LexA is selfregulated but also rapidly degraded while the SOS-inducing signal is active, a process that overcomes the increased expression due to derepression. It is also consistent with these proposals that SOS gene induction occurs late rather than early in induction. Several other sets of up-regulated genes presumably respond to the altered cellular conditions of phage growth. Induction of subunits of ribonucleotide reductase genes may reflect increased need for dNTPs. The set of genes involved in phosphate metabolism and transport responds to cellular phosphate limitation, which may occur in the induced cell; the involvement of pho genes in lambda growth has been noted (25). The genes cpxP, htpX, and degP are activated by the two component system encoded by cpxR/ A, which senses envelope stress (26), a plausible response to such phage proteins as the lysis set that are made at late times of induction. Liu et al.
Summary and Further Discussion Ribosome profiling provides a detailed view of gene expression of both phage and host during the development of bacteriophage lambda. This analysis confirms the expected general patterns of gene expression but also shows an unexpected complexity of the translation landscape. The major finding is that much ribosome occupancy occurs in ORFs of unknown function. Although some of these potential ORFs were known from analysis of the lambda genome sequence, and all could of course be inferred by simple analysis, the translation data focuses attention on a substantial number that were not previously known to be of interest. Numerous translated ORFs are large enough plausibly to encode functional proteins, in the range of 20–130 aa. The underappreciated importance of small proteins has been recognized, and they have a variety of roles that might be relevant to lysogen and phage growth, acting as intercell signaling factors, toxins, and membrane components (27). Few of the newly discovered ORFs appear to be expressed in the lysogen at levels similar to well known prophage-specific genes (cI, rexA, rexB, lom, bor), although the previously recognized ORFs ea59 and ea8.5 are significantly expressed. However, very low expression might well be important. Of course, the act of ribosome engagement or ribosome synthetic activity could be important rather than the translation 1. Hendrix RW, Roberts JW, Stahl FW, Weisberg RA (1983) Lambda II (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY). 2. Hendrix RW, Casjens S (2006) Bacteriophage lambda and its genetic neighborhood. The Bacteriophages, ed Calendar RL (Oxford Univ Press, Oxford). 3. Chen Y, Golding I, Sawai S, Guo L, Cox EC (2005) Population fitness and the regulation of Escherichia coli genes by bacterial viruses. PLoS Biol 3(7):e229. 4. Osterhout RE, Figueroa IA, Keasling JD, Arkin AP (2007) Global analysis of host response to induction of a latent bacteriophage. BMC Microbiol 7:82. 5. Ingolia NT (2010) Genome-wide translational profiling by ribosome footprinting. Methods Enzymol 470:119–142. 6. Ingolia NT, Brar GA, Rouskin S, McGeachy AM, Weissman JS (2012) The ribosome profiling strategy for monitoring translation in vivo by deep sequencing of ribosomeprotected mRNA fragments. Nat Protoc 7(8):1534–1550. 7. Stern-Ginossar N, et al. (2012) Decoding human cytomegalovirus. Science 338(6110): 1088–1093. 8. Oh E, et al. (2011) Selective ribosome profiling reveals the cotranslational chaperone action of trigger factor in vivo. Cell 147(6):1295–1308. 9. Stein LD, et al. (2002) The generic genome browser: A building block for a model organism system database. Genome Res 12(10):1599–1610. 10. Li GW, Oh E, Weissman JS (2012) The anti-Shine-Dalgarno sequence drives translational pausing and codon choice in bacteria. Nature 484(7395):538–541. 11. Oppenheim AB, Kobiler O, Stavans J, Court DL, Adhya S (2005) Switches in bacteriophage lambda development. Annu Rev Genet 39:409–429. 12. Friedman DI, Gottesman M. (1983) Lytic mode of lambda development. Lambda II, ed Hendrix RW, Roberts JW, Stahl FW, Weisberg RA (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY). 13. Gussin G, Johnson A, Pabo C, Sauer R (1983) Repressor and cro protein: Structure, function, and role in lysogenization. Lambda II, eds Hendrix RW, Roberts JW, Stahl FW, Weisberg RA (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY). 14. Roberts JW, et al. (1998) Antitermination by bacteriophage lambda Q protein. Cold Spring Harb Symp Quant Biol 63:319–325. 15. Ray PN, Pearson ML (1974) Evidence for post-transcriptional control of the morphogenetic genes of bacteriophage lambda. J Mol Biol 85(1):163–175. 16. Sampson LL, Hendrix RW, Huang WM, Casjens SR (1988) Translation initiation controls the relative rates of expression of the bacteriophage lambda late genes. Proc Natl Acad Sci USA 85(15):5439–5443.
Liu et al.
product itself. Thus, translation of ORFs upstream of expressed genes has a well-known regulatory function in yeast (28), and the very act of ribosome binding can change the structure and availability of mRNA for translation of other parts of the message, or for other functions of RNA such as transcription termination. A specific example of an alternate consequence of ribosome function is the expression of the lambda bar “genes,” which act to sequester ribosomes that must be freed by a peptide hydrolase (29). Another possibility is that engagement of RNA with ribosomes serves to prevent deleterious activity (e.g., forming R loops) of free RNA (30), which might be prevalent where antiterminators prevent Rho function, as in most lambda transcription; thus, functional ORFs might be retained in regions of the genome where transcription occurs, even if no functional polypeptide is produced. Finally, it is of course possible that some or all of these ORFs have no function at all, representing just background activity of the transcription and translation systems. We cannot eliminate the possibility that there is adventitious and nonfunctional association of RNA with ribosomes that gives rise to illusory translation of at least some of these potential ORFs. Further informatic and directed analysis, and comparative phage genomic analysis, would be required to query the function of this unexpected translation activity. Materials and Methods The strain used in these experiments was E. coli K12 MG1655 (obtained from F. Blattner, University of Wisconsin, Madison, WI), made lysogenic for λcI857 (from M. Gottesman, Columbia University Medical School, New York). Ribosome protected RNAs were prepared as described (6), with minor modifications. Detailed protocols and methods of data analysis are provided in SI Materials and Methods. ACKNOWLEDGMENTS. We thank Ryland Young and Jim Hu (Texas A&M University) for careful and thoughtful reading of the manuscript, Sherwood Casjens (University of Utah) for helpful comments, and Nick Ingolia for providing a protocol before publication. Help with the display and analysis of the data was provided by the PortEco project through National Institute of General Medical Sciences U24GM088849, and this work was supported by National Institutes of Health Grant GM 21941.
17. Dai XX (2009) Expression, purification and characterication of bacteriophage lambda tail tip proteins. PhD thesis (Univ of Pittsburgh, Pittsburgh). 18. Parma DH, et al. (1992) The Rex system of bacteriophage lambda: Tolerance and altruistic cell death. Genes Dev 6(3):497–510. 19. Xu J, Hendrix RW, Duda RL (2004) Conserved translational frameshift in dsDNA bacteriophage tail assembly genes. Mol Cell 16(1):11–21. 20. Henkin TM, Yanofsky C (2002) Regulation by transcription attenuation in bacteria: How RNA provides instructions for transcription termination/antitermination decisions. Bioessays 24(8):700–707. 21. Juhala RJ, et al. (2000) Genomic sequences of bacteriophages HK97 and HK022: Pervasive genetic mosaicism in the lambdoid bacteriophages. J Mol Biol 299(1):27–51. 22. Hendrix RW, Duda RL (1992) Bacteriophage lambda PaPa: Not the mother of all lambda phages. Science 258(5085):1145–1148. 23. Sergueev K, Court D, Reaves L, Austin S (2002) E.coli cell-cycle regulation by bacteriophage lambda. J Mol Biol 324(2):297–307. 24. Little JW, Mount DW (1982) The SOS regulatory system of Escherichia coli. Cell 29(1): 11–22. 25. Maynard ND, et al. (2010) A forward-genetic screen and dynamic analysis of lambda phage host-dependencies reveals an extensive interaction network and a new antiviral strategy. PLoS Genet 6(7):e1001017. 26. Vogt SL, Raivio TL (2012) Just scratching the surface: An expanding view of the Cpx envelope stress response. FEMS Microbiol Lett 326(1):2–11. 27. Hobbs EC, Fontaine F, Yin X, Storz G (2011) An expanding universe of small proteins. Curr Opin Microbiol 14(2):167–173. 28. Hinnebusch AG (2005) Translational regulation of GCN4 and the general amino acid control of yeast. Annu Rev Microbiol 59:407–450. 29. Ontiveros C, Valadez JG, Hernández J, Guarneros G (1997) Inhibition of Escherichia coli protein synthesis by abortive translation of phage lambda minigenes. J Mol Biol 269(2):167–175. 30. Leela JK, Syeda AH, Anupama K, Gowrishankar J (2013) Rho-dependent transcription termination is essential to prevent excessive genome-wide R-loops in Escherichia coli. Proc Natl Acad Sci USA 110(1):258–263. 31. Ingolia NT, Ghaemmaghami S, Newman JR, Weissman JS (2009) Genome-wide analysis in vivo of translation with nucleotide resolution using ribosome profiling. Science 324(5924):218–223.
PNAS | July 16, 2013 | vol. 110 | no. 29 | 11933
GENETICS
Finally, there is a set of up-regulated genes that mostly have known function, but no obvious relation to phage development; examples are shown in Fig. 5C. Phage growth substantially engages the protein production capacity of the cell. At 20 min after induction, about 30% of the total ribosome-protected RNA reads map to the lambda genome. Of the 1,000 down-regulated genes (Datasets S4 and S5), 600 are downregulated more than twofold. There appears to be a constancy of total protein synthetic capacity when the number of ribosome reads from phage and host are summed throughout the 20-min period of phage development; the 30% increase in phage reads is matched by a 30% loss in host reads (Fig. 5 D and E), and the decrease in each individual amino acid incorporation into bacterial protein is equal to its incorporation into phage proteins (Fig. S5).